Deepfake Detection on FakeAVCeleb (test) benchmark leaderboard · SOTA2 Research

Deepfake Detection on FakeAVCeleb (test)

99.8AccuracyHAVIC

Evaluation Results

Method	Links
HAVIC2026.03		99.8	—	—	—	99.9
FGMDF2026.04		99.46	93.68	90.82	92.23	99.98
FGMDF2026.04		99.46	100	96.97	98.46	99.98
FGMDF2026.04		99.46	99.52	99.68	99.6	99.98
FGMDF2026.04		99.46	100	100	99.9	99.98
AVTENet_ff2023.10		99	100	97	99	—
AVTENet_ff2023.10		99	97	100	99	—
MSCT2026.04		98.75	—	—	—	98.83
PIA2026.03		98.7	—	—	—	99.8
AVFF2024.06		98.6	—	—	—	99.1
AVFF2026.03		98.6	—	—	—	99.1
BusterX2026.04		96.3	—	—	—	—
MIS-AVioDD2023.10		96	—	—	—	—
FGI2026.04		95.15	64.63	76.81	70.19	98.98
FGI2026.04		95.15	93.48	97.9	95.64	98.98
FGI2026.04		95.15	80	85.71	82.75	98.98
FGI2026.04		95.15	99.3	94.01	96.58	98.98
MRDF-CE2026.03		94.1	—	—	—	92.4
MRDF-CE2026.04		94.05	—	—	—	92.43
AV-Lip-Sync2023.10		94	93	96	94	—
AV-Lip-Sync2023.10		94	96	93	94	—
Multimodaltrace2023.10		93	—	—	—	—
AVFakeNet2023.10		93	—	—	—	—
RealForensics2024.06		89.9	—	—	—	94.6
RealForensics2026.03		89.9	—	—	—	94.6
Ensemble2023.10		89	83	99	90	—
Ensemble2023.10		89	98	80	88	—
AVDF2026.04		87	81.22	89.75	85.27	97.47
AVDF2026.04		87	88.02	79	83.27	97.47
AVDF2026.04		87	87.47	92.5	89.91	97.47
AVDF2026.04		87	92.29	86.75	89.43	97.47
AVoiD-DF2023.10		84	—	—	—	—
AVoiD-DF2024.06		83.7	—	—	—	89.2
AVoiD-DF2026.03		83.7	—	—	—	89.2
AVOID-DF2026.04		83.7	—	—	—	89.2
MDS2024.06		82.8	—	—	—	86.5
MDS2026.04		82.8	—	—	—	86.5
VFD2026.04		81.52	—	—	—	86.11
VFD2024.06		81.5	—	—	—	86.1
VGG162023.10		81	69	90	78	—
VGG162023.10		81	87	78	82	—
LipForensics2024.06		80.1	—	—	—	82.4
LipForensics2026.03		80.1	—	—	—	82.4
AVFakeNet2024.06		78.4	—	—	—	83.4
Emotions Don’t Lie2024.06		78.1	—	—	—	79.8
VGG16 (Soft-Voting)2023.10		78	69	90	78	—
VGG16 (Soft-Voting)2023.10		78	90	69	78	—
VGG16 (Hard-Voting)2023.10		78	69	90	78	—
VGG16 (Hard-Voting)2023.10		78	90	69	78	—
Xception2023.10		76	88	61	72	—
Xception2023.10		76	70	91	80	—
LipForensics2023.10		76	70	91	80	—
LipForensics2023.10		76	88	61	72	—
CViT2024.06		69.7	—	—	—	71.8
Not-made-for-each-other2023.10		69	62	99	76	—
Not-made-for-each-other2023.10		69	94	40	57	—
Xception2024.06		67.9	—	—	—	70.5
Xception2026.03		67.9	—	—	—	70.5
Multimodal-22023.10		67	71	59	64	—
Multimodal-22023.10		67	65	76	70	—
FTCN2024.06		64.9	—	—	—	84
FTCN2026.03		64.9	—	—	—	84
CDCN2023.10		52	50	7	12	—
CDCN2023.10		52	50	94	65	—
Multimodal-12023.10		50	0	0	0	—
Multimodal-12023.10		50	50	100	66	—